Wednesday, December 22, 2010

A Few Follow-up Thoughts on the Falcons Post

  • It's worth mentioning that a low variance offense can be very effective, even though it ends up with a lower yards per play number. Georgia Tech's 2009 ACC Championship team is a perfect example. Tech averaged 6.2 yards per play, which is good, but only ranked 17th in the country. According to Football Outsiders' FEI rankings, however, Tech had the best offense in the country. Why? Because FEI is drive-based and Tech's offense was excellent at scoring by consistently stringing together five-yard gains and by avoiding long yardage situations that kills drives. All things considered, you would rather have an offense that gains five yards on every play than one that gains seven yards per play on two 35-yard plays and eight plays stopped for no gain. The Falcons seem to follow that model. The Eagles are the polar opposite; they have an offense that generates a high yards per play number on the strength of a wealth of big plays, but they are also high variance, which is how they found themselves down 31-10 to the Giants with eight minutes to go on Sunday.

  • A second benefit of a low variance offense that strings together consistent, medium-sized gains is that a team with that sort of offense will hold onto leads very well. Remember the stat during the Cowher years about how the Steelers almost never lost games in which they led in the fourth quarter? Mike Mularkey has brought that tendency to Atlanta. Here's your stat of the day: the Falcons have not lost a game in which they held a fourth quarter lead since the road game against the Saints in 2008. The Smith/Mularkey regime has only lost two times when the Falcons had a lead in the fourth quarter: the games against New Orleans and Denver in 2008. For Falcons fans like me whose formative memory is the home playoff loss to Dallas in 1980, this is a refreshing feeling. It's like rooting for a team with a great bullpen. If the Falcons have a lead heading into the fourth quarter, they're in great shape, even with a suspect pass defense.

  • Watching Michigan on Saturdays and then the Falcons on Sundays has been an Elvis-like experience of taking uppers and downers. Even with an appalling defense, Michigan had a reasonably good yards per play margin this year because the offense averaged 6.9 yards per play, which was 7th nationally. However, Michigan's yardage advantages didn't translate onto the scoreboard because of what I described as the "dumb shit" factor: red zone turnovers, missed field goals, and terrible kickoff coverage and punt returns that gave opponents field position advantages. The Falcons are 180 degrees from that. The Falcons are outgained by a significant margin on a per-play basis, but they get the dumb shit. They don't turn the ball over, they make field goals, they don't commit penalties, and they are good on special teams. If "dumb shit" is a way to measure the effect of coaching, then Mike Smith should be coach of the year and Rich Rodriguez should be making way for Jim Harbaugh.

  • When I did the chart of conference champions from the pas decade, 2008 stood out to me for a couple reasons. First, Arizona was better than their record. The '08 Cardinals were not strong in terms of points and Football Outsiders was very down on them, but their yardage margin was better than the margins of a number of Super Bowl champions (admittedly against a weak schedule). It shouldn't have been a massive surprise that they made the Super Bowl. The team they played in the Super Bowl had the best defense on a yards per play basis of any Super Bowl team in the decade. The '08 Steelers were three-tenths of a yard better than the '02 Bucs and four-tenths of a yard better than the team that struck me as having the best defense at least since the '85 Bears: the '00 Ravens. (The '85 Bears allowed 4.4 yards per play, which is surprisingly high, although that team did force a ridiculous 52 turnovers. The '76 Steelers - the Steel Curtain at its best - allowed 3.8 yards per play.) Given the quality of the teams in the game and the way that it played out, Super Bowl XLIII deserves consideration as one of the best Super Bowls of all time.

  • One argument in favor of the way that championships are awarded in college football: the two best teams of the past decade in the NFL were the '01 Rams and the '07 Patriots and neither team won the Super Bowl. In both instances, those teams lost the Super Bowl to opponents with demonstrably inferior records whom the '01 Rams and '07 Pats had beaten on the road during the regular season. It strains the meaning of "champion" to assign that title to the '01 Patriots and '07 Giants. I remain in favor of a small college football playoff; four teams would be good, six teams would be very good, and eight teams would be OK. However, we shouldn't ignore the fact that the absence of the playoff reset button in college football is a good thing in a significant way.


Anonymous said...

Great post! Several thoughts and disagreements:

1) I don't think low variance offense is a viable strategy. I think it puts a team in coinflip games, and that eventually over time, those flips even out. Carr "tried" low variance style football, and he constantly lost 4th quarter leads, and Carr's teams were perpetually playing coinflip games against overmatched opponents.

2) FEI and other drive based (aka "Hidden game") stats have less predictive validity than simple ole' Yards Per Play.

3) Though the 01 Rams and 07 Pats might have deserved to be champions, the playoff system that knocked each team off was very entertaining. Further, both teams beat the eventual champs by roughly one score, so its pretty tough to unequivocally state that one team was so much "better" than the other that the other didn't deserve to be considered a champion.

4) The Eagles were losing to the Giants because of turnovers. Once they stopped turning the ball over, their offense dominated the game.

Michael said...

1. Low variance is a good strategy if a team can pull it off. It's hard as hell to avoid negative plays and consistently churn out medium-sized gains. The Falcons are able to do it, but I don't remember many other teams pulling it off. (The '90 Giants and several Steelers teams come to mind.) The problem with Lloyd's teams is that they weren't really low variance. A low variance offense can run the same plays throughout the game and consistently churn out long drives. Lloyd, on the other hand, had two offenses: the scoring offense and the run-the-clock-out offense. The scoring offense was often excellent, but not in a low variance way. With all of the big plays that Michigan hit in '99 and '03 (to use two years that could have been better without blown leads), those offenses were anything but low variance. The problem arose when Lloyd would change the offense and become uber-predictable with a lead. In other words, he didn't play to his teams' strengths when he had great offenses. That's a different problem.

2. Do you have a link on that? I love any support for the idea that YPP is the best way to measure a team.

3. A one-score win at home is one thing, but a one-score win on the road is a reasonably definitive result. I agree that the NFL playoffs are exciting, but the regular season is devalued by knowing that one team can be far superior over 16 games and still lose out to an inferior opponent based on one game. (Yes, I know that based on the TV ratings, I am on an island on this topic.)

4. The Eagles ran 22 plays in the first half. Only two of their six drives were stopped by turnovers. I think the better conclusion is that their offense was terrible in the first half.